BMT-capped HgTe QDs exhibited a distinct exciton feature at approximately 2300 nm, with absorption extending up to 3000 nm (Fig. 1a). After five cycles of spin casting, a uniform QD film with a surface roughness (R_a) of 8.99 nm and a thickness of around 80 nm was obtained. CdS was selected as the capping layer of the QD film due to its matching energy levels [11] and the strong bonding affinity between sulfur (S) and mercury (Hg). Traditional CBD typically requires an aqueous environment and elevated temperatures around 80 °C, which can lead to degradation and sintering of HgTe QDs [22–24]. To address these challenges, a modified CBD approach was developed in this work (Fig. 1b). By using thioacetamide as the sulfur precursor, the deposition process was carried out at a reduced temperature, minimizing the risk of QD degradation while ensuring uniform and conformal layer formation. Although the CBD method may result in lower CdS crystallinity and stoichiometric composition purity, these limitations can be mitigated by careful device design, as will be elaborated later. All reactions were conducted in an anhydrous solvent. We optimized the deposition conditions by tuning the temperature range and reaction time, identifying 55 °C for 10 min as the optimal parameters. At this temperature, sulfur is efficiently released from thioacetamide, while the reaction between cadmium and sulfur—regulated by triethylamine—proceeds gradually [27]. This controlled process ensures uniform layer formation while releasing sufficient sulfur to interact effectively with the underlying HgTe QD layer, as will be discussed in detail later. The optimized CBD process yields a compact CdS film while preserving the structural integrity of the underlying HgTe QD layer, as evidenced by AFM images of the QD thin films before and after CBD (Fig. 1c). The height profiles (Figure S1) of both the pristine HgTe QD film and CdS/HgTe heterostructure reveal an increase in film thickness with averaged values from 80 ± 5 nm to 90 ± 5 nm after CdS deposition obtained from multiple measurements across the active layer/substrate edge. This indicates partial infiltration of CdS into the HgTe QD layer and the formation of a thin interfacial layer. A cross-sectional SEM image of the CdS/HgTe QD heterojunction deposited on a silicon wafer (Fig. 1d) show that the HgTe QDs form a continuous and uniform layer, while the CdS layer exhibits clear phase separation from the underlying HgTe layer. Furthermore, as shown in Figure S2, the absorption edge of the deposited CdS film occurs near 430 nm, significantly shifted from that of bulk CdS. The pronounced absorption tail extending up to 600 nm can be ascribed to the presence of nanocrystal aggregates and light scattering phenomena within the CdS thin film. Additionally, this extended absorption may reflect the presence of a significant number of defect states in the CdS layer. These defect states may act as long-lived traps, which beneficially modulate the photoconductive gain of the CdS/HgTe photodetectors. Next, we investigated the interface of the CdS/HgTe QD layers. Raman spectroscopy was employed to compare the HgTe QD films before and after CBD treatment. As shown in Fig. 2a, the pristine HgTe QD film exhibits four distinct peaks, corresponding to the vibrational modes of acoustical and optical phonons in HgTe QD lattice [11]. We also emphasize the presence of Hg-S chemical bonds in the QD film which is attributed to the strong attachment of BMT ligand via the thiol functional group to the Hg-rich QD surface. One has to admit that due to the similar bond lengths between in Hg-Te and Hg-S (around 2.8 Å) it is challenging to distinguish the latter in the structure of LO2 peak at 260 cm^{− 1} of Raman spectrum of HgTe QDs [28]. After the CBD process, a new Raman peak emerges at 305 cm^{− 1}, which is characteristic of CdS, accompanied by a notable increase in the Hg-S signal from 200 to 270 cm^{− 1} and a concurrent reduction in the HgTe peak intensity [29]. These spectral changes, specifically the emergence of the bands at 205, 235 and 257 cm^{− 1}, correlate well with the literature data on multi phonon scattering in nanocrystalline CdS [30, 31]. Moreover, the broadening of the bands at 250–260 cm^{− 1} suggests that sulfur ions from the bath have infiltrated the HgTe film, resulting in substantial interfacial modification. This conclusion is further supported by depth-profiled XPS conducted via time-resolved argon ion etching (Fig. 2b). The Cd and S signals are predominantly detected at the surface; the Cd signal rapidly attenuates with increasing etching time, while the S signal exhibits a more dispersed profile, extending deeper into the film. In contrast, the Hg signal is present from the surface, whereas the Te signal is initially weak near the surface but intensifies with depth. These contrasting depth-dependent trends reveal a clear spatial separation of elements, confirming that the CdS layer is primarily located on the top surface, while a portion of Hg-S also forms at the interface, contributing to the interfacial modification of the underlying HgTe QDs. This conclusion is further supported by XPS comparative analysis on the bare HgTe QD film, CdS film, as well as their heterojunction structure. As shown in Fig. 2c, the binding energies of the Hg 4f, Te 3d, and Cd 3d peaks remain unchanged after hybridization, indicating that the CBD of CdS does not change the local chemical environments of Hg and Te [17]. The consistent and weak intensity of oxidized Te component (Te-O) suggests the resistance of HgTe QDs to oxidation during CBD and device fabrication. The high resolution XPS of HgTe evidences on the presence of two Hg species: core Hg bound to Te (Hg-Te) and surface Hg bound to thiol-containing ligands (Hg-S). The increased relative area of the Hg-S component in the hybrid CdS/HgTe structure indicates effective passivation of free surface Hg ions by sulfur ions during CBD. The S 2p spectrum exhibits a more complex profile combining different sulfur species. The fitting of XPS data on CdS film evidences on the presence of S-Cd (S bound to Cd, S 2p_3/2 at 161.7 eV) [32] and organic sulfur compound S-C (S bound to C, S 2p_3/2 at 170.3 eV) associated with unreacted thioacetamide. The HgTe sample shows a signal (S 2p_3/2 at 162.2 eV) corresponding to sulfur from thiol groups attached to mercury (Hg-S). A broad and intense peak at 169 eV corresponds to the Te 4s signal in the HgTe sample. The sulfur signal in the CdS/HgTe heterojunction structure represents a combination of all mentioned sulfur species. A subtle peak shift of Hg-S component confirms that unpassivated Hg²⁺ on the HgTe surface bond to S²⁻ during the CBD, effectively eliminating dangling bonds. Figure 2b illustrates such composition gradient at the CBD-processed CdS/HgTe QD interface. To investigate the energy-level alignment at the interface, UPS was conducted on bare HgTe and CdS films, as presented in Fig. 3a. The valence band maxima of HgTe and CdS were found to lie 0.10 and 2.28 eV below the Fermi level, respectively. The work functions of HgTe and CdS were calculated to be 4.44 and 4.37 eV, respectively. Based on the bandgap value calculated from the absorption spectrum, the conduction band minimum values are estimated to be 4.13 and 4.25 eV for HgTe and CdS, respectively. Combining the band information of each material, the schematic diagram of the heterojunction is provided in Fig. 3b, showing a typical type-II p-n heterojunction. The energy levels of the Hg-S bonded interfacial region are taken from the literature [33, 34]; however, due to the ultrathin, and likely non-crystalline nature of this layer, its role in modulating electron extraction is expected to be insignificant. In this configuration, electrons are driven into the CdS electron accepting layer, while holes are blocked by the interface energy barrier and transported along the source–drain channel toward the external circuit, resulting in an enhanced photocurrent. Utilizing the CdS/HgTe QD heterostructure, we fabricated a phototransistor, as illustrated in Fig. 3c and Figure S4. For comparative analysis of the device operation mechanism, a control phototransistor based on a bare HgTe QD layer was also fabricated under identical fabrication conditions. The typical transfer characteristics of the HgTe and CdS/HgTe phototransistors under dark conditions are shown in Fig. 3c. The fabricated pristine HgTe QDs-based phototransistor exhibits a typical ambipolar field-effect behavior with hole and electron mobility of 0.09 and 0.006 cm² V^{− 1} s^{− 1}, respectively [35], and on/off current ratio ≈ 11. Upon introducing CdS layer, the CdS/HgTe heterojunction device demonstrates an enhanced conductivity with p-type mobility of 0.445 cm² V^{− 1} s^{− 1} and an improved on/off ratio ≈ 22. The transition from ambipolar to unipolar transistor behavior, along with a doubled on/off ratio, confirms effective electron transfer from the HgTe QD layer to the CdS layer. This transfer suppresses electron transport in the HgTe QD layer (eliminating n-type behavior) and reduces the recombination current, thereby lowering the dark current. Meanwhile, the significantly increased hole mobility indicates a substantial reduction in trap states, likely resulting from defect passivation induced during the CBD process. Interestingly, when the cadmium precursor is removed from the CBD process, leaving only the sulfur precursor, the post-treated HgTe transistor still exhibits the same level of hole mobility as the CdS/HgTe transistor (Figure S5). This observation suggests that the dominant mechanism for the improvement in hole transport is sulfur-induced defect passivation rather than the intrinsic properties of the CdS layer. A detailed comparison of the hysteresis and gate leakage currents is shown in Figure S6 of the Supporting Information. The transfer curves show that the pristine HgTe transistors exhibit strong hysteresis, indicating a high density of slow trap states at the surface and interfaces. After CdS deposition by CBD, the hysteresis is reduced, suggesting that CdS effectively passivates the HgTe QDs' surface. Next, we evaluated the photodetection performances of the devices. As shown in Fig. 4a, the responsivity measured at a gate voltage of 5 V under 1550 nm laser illumination with a power density of 11 mW cm^− 2 is significantly enhanced from 0.2 A W^− 1 for the pristine HgTe device to 1.3 A W^− 1 for the CdS/HgTe device. This improvement is consistent with the increased hole mobility and the efficient separation of electrons and holes facilitated by the heterostructure. As shown in Figure S5, only sulfur-containing groups were introduced to modify the QD layer properties. The S-only/HgTe device exhibits improved hole transport, as evidenced by the p-type current reaching a level comparable to that of the CdS/HgTe device. In contrast, the modification of the n-type regime is much weaker in the S-only/HgTe device than in the CdS/HgTe device. Moreover, the responsivity of the S-only/HgTe device is significantly lower than that of the CdS/HgTe device. These results demonstrate that Cd is critical for forming the CdS electron-accepting layer, which efficiently channels electrons away from holes and enhances photoconductive gain, thereby clarifying the importance of the Cd component in the modification. Furthermore, the performance gap between the two devices, shown in Fig. 4b, becomes even more pronounced at lower light intensities. At a light intensity of 1.1 µW cm^− 2, the responsivity increases from 13.33 A W^− 1 for the pristine HgTe device to 365.51 A W^− 1 for the CdS/HgTe device. The corresponding external quantum efficiency (EQE), estimated using the relation EQE = (1240 W nm A^{− 1}) R/λ is as high as ~ 29,200%, which clearly demonstrates a pronounced photoconductive gain. The photoconductive gain, arises from the efficient trap passivation and separation of electrons and holes within the CdS/HgTe heterostructure, particularly benefiting responsivity enhancement at low light intensities. Fig. 4c shows the photocurrent (I_ph) as a function of light intensity (I), following a power-law dependence described by I_ph∝I^α. For both devices under high light intensities, the fitted exponent α approaches 0.5, suggesting that bimolecular recombination is the dominant loss mechanism under illumination [36]. However, at low light intensities the pristine HgTe device shows α >1, indicating the dominance of trap-assisted recombination under weak illumination [26, 36, 37]. The absence of the such regime in the CdS/HgTe device suggests the effective reduction of trap states in the QD layer.

Fig. 4d presents the current noise spectral density profiles for the devices. Both devices exhibit a distinct 1/f trend, characteristic of dominant flicker noise. Such noise is commonly observed in thin film transistors at low frequency and can be attributed to fluctuation in mobile carrier density and mobility. Compared to the pristine HgTe device, the CdS/HgTe structure exhibits a significant reduction in noise levels. Notably, under positive gate voltage (V_GS = 5), the noise density is further suppressed, highlighting the superior low-noise characteristics enabled by the interfacial modification and heterostructure. Owing to the low current noise, the detectivity (D^*) of the CdS/HgTe heterojunction device reaches 4.43 × 10¹¹ Jones at an illumination wavelength of 1550 nm under low illumination intensity (1.1 µW cm^− 2) at a frequency of 2.5k Hz, which is significantly better than that of the pristine HgTe QD device (1.36 × 10⁹ Jones). Such remarkable increase in detectivity results from the simultaneous enhancement of responsivity and the reduction of noise. Fig. 4e presents the wavelength-dependent specific detectivity (D^*) spectrum of the CdS/HgTe phototransistor. The device exhibits D^* of around 10¹⁰ Jones at a wavelength of 2500 nm with a power density of 0.34 mW cm^− 2. Notably, the device maintains high sensitivity (D^* >10¹⁰ Jones) at room temperature across an extended spectral range up to 2500 nm.

The transient photoresponses of both devices provide further insight into the effects of interfacial modification, as illustrated in Fig. 4f and Figure S8. The decay dynamics of the normalized photocurrent were recorded following the cessation of illumination, and the effective decay time constant (τ_eff) was extracted [38–41]. Upon the interfacial modification and the formation of the heterostructure, an additional fast decay component (~ 28.5 µs) emerges, which we attribute to the passivation of HgTe QD surface defects, which facilitates efficient charge transport and promotes bimolecular recombination. Meanwhile, the slower decay component (~ 195 µs) is significantly prolonged compared to pristine HgTe (~ 99.2 µs). This originates from long-lived traps, which may reside in the CdS layer, that sustain the photoconductive gain, whereas in the unmodified HgTe QDs-based device, a single trap-limited bulk recombination mechanism dominates [42–44]. These findings further highlight the role of sulfur ion infiltration during the CBD process in partially passivating surface trap states of the HgTe QDs and thus improving the response speed of the photodetector.